Skip to main content

Assimilating Models of Designers, Users and Systems

Objective

The aims of AMODEUS were to:
-develop formal models of the behaviour of interactive systems as a means of formulating claims about usability and for generating interactive software architectures
-formulate explicit cognitive models of user behaviour and to represent them in a manner that enables cognitive theory to be migrated into design practice
-study how interface designers reason about interface properties and usability
-develop a notation, called a design rationale, to represent design options and their properties
-explore inter-relationships among conceptions of user behaviour, system behaviour and designs, and to examine their integration.
Interdisciplinary approaches to the study of user/system interactions were developed using cognitive models of users, formal representations of system behaviour, and their refinement into interactive system architectures. Interface designers were studied and representations developed to support design decision making.

The system modelling activity was extended significantly to include graphical and cooperative systems. An agent notation was developed that draws together the basic theoretical models with the practical requirements of object oriented interactive system design. The interactive architectures work made significant progress towards heuristics for agent based design of interactive systems. An expert system embodying such heuristics is 1 of several proof of concept demonstrations.

Work in both forms of user modelling has significantly extended the scope of available techniques. In one approach, technique based around the concept of the designer's intended procedure has also been developed where the instruction language for a programmable user architecture is interactively applied and run within the architecture to reveal behavioural consequences of a design. In the other approach a proof of concept expert system has been developed to model the cognitive activity of users and provide design advice across a significant range of topics.

The design rationale developed a particular representation for supporting design activity. Many design scenarios have now been analysed and the representation has been used to integrate over user, system and other criteria.

Cognitive and system modellers have continued to develop an interaction framework for synthesizing over areas, and a number of other notational frameworks and ideas for interdisciplinary integration were developed.

The results of the system architecture modelling activity performed in the context of the project have been assessed. The goal of this research package is to provide insights and effective guidelines for designing and implementing interactive software. Although multiagent models are considered to be a promising way of supporting human factors as well as software engineering requirements, they are hard to apply. The description of the overall architecture of an interactive system has been clarified, making explicit the boundaries with the functional core and the presentation toolkit, two components considered by programmers as unavoidable constraints. The dialogue controller has been refined into a number of agents modelled along two dimensions: facets and levels of abstractions. The combination of these two axes is a convenient way of describing phenomena observed from current practices. Several heuristic rules have been identified, which provide useful insights and guidance for the definition of relationships between agents.

The goal of the project is to develop interdisciplinary approaches to studying human computer interaction and to move towards applying the results to the practicalities of design. One of the approaches the project is taking to represent design is design space analysis. One of its goals is to help bridge the gap from relatively theoretical concerns to the practicalities of design. Design space analysis is a central component of a framework for representing the design rationale for designed artifacts. The current work focuses more specifically on the design of user interfaces. A design space analysis is represented using the QOC notation, which consists of questions identifying key design issues, options providing possible answers to the questions, and criteria for assessing and comparing the options. An overview of the approach is given. Some examples of the research issues currently under consideration are provided.
APPROACH AND METHODS
The research was an inter-disciplinary collaboration which drew upon concepts and tools from software engineering, cognitive psychology, and artificial intelligence.
Within the system modelling research, emphasis was upon the precise specification of abstract properties governing interactions and on the means of inter-relating and refining such properties into actual interactive software architectures. Exemplar systems provided the material for analysis. User modelling explored techniques for representing cognitive theory in the context of design. One technique involved the development of an instruction language for a programmable user model, the other required the develoment of an expert system that reasons about the properties human cognition in user-system interaction. Both experimental evidence and evidence drawn from scenarios of user behaviour were used to support theory development.
The design rationale work describes and specifies options within a space of possible designs. Both system and user properties are represented within this technique, as are the criteria for choosing among options. As with the system modelling work, different exemplars were analysed. Using empirical protocol and questionnaire techniques, the Action also studied how designers reason about interfaces in commercial design projects.
The project was integrated through the use of a structured methodology in which researchers from different perspectives provided analyses of a common and systematically sampled set of scenarios describing user system exchanges and design options. The input from the different perspectives was analysed, contrasted and integrated.
PROGRESS AND RESULTS
The results of the complete Action are described in 23 deliverables. Over the course of the project:
The scope of the Action's system modelling activity was extended significantly to include graphical and co-operative systems. An agent notation was developed that draws together the basic theoretical models with the practical requirements of object oriented interactive system design. The interactive architectures work made significant progress towards heuristics for agent based design of interactive systems. An expert system embodying such heuristics is one of several proof-of-concept demonstrations resulting from the Action.
Work in both forms of user modelling has significantly extended the scope of our available techniques. In one approach, technique based around the concept of "the designer's intended procedure" has also been developed. In this technique, the instruction language for a programmable user architecture is interactively applied and run within the architecture to reveal behavioural consequences of a design. In the other approach a proof-of-concept expert system has been developed to model the cognitive activityof users and provide design advice across a significant range of topics.
The design rationale developed a particular representation for supporting design activity. Many design scenarios have now been analysed and the representation has been used to integrate over user, system and other criteria. The design related research hasalso reported on a number of laboratory and field studies in which real designers deal with their own designs or discuss the same scenario material that has been analysed by the user and system modellers within the Action.
With respect to the integration of ideas from the different perspectives within the project, three sets of communal analyses were carried out. Each approach within the project analysed the same set of basic material. Cognitive and system modellers have continued to develop an "interaction framework" for synthesising over areas, and a number of other notational frameworks and ideas for interdisciplinary integration were developed.
POTENTIAL
The research has forced significant extensions to the scope of our modelling techniques. By relating these developments directly to design practice and providing a number of proof of concept demonstrations, the project should have provided the wider HCI community with a firm foundation to build systematic bridges both between different theoretical approaches and from these approaches to the realities of commercial design practise.

Coordinator

MRC Applied Psychology Unit
Address
15 Chaucer Road
CB2 2EF Cambridge
United Kingdom

Participants (6)

Alcatel SEL AG
Germany
Address
Lorenzstraße 10
70435 Stuttgart
KOBENHAVNS UNIVERSITET
Denmark
Address
Njalsgade, 94
2300 Copenhagen
Logica Ltd
United Kingdom
Address
64-68 Newman Street
W1A 4SE London
Rank Xerox EuroParc Ltd
United Kingdom
Address
61 Regent Street
CB2 1AB Cambridge
University of York
United Kingdom
Address
Heslington -
YO1 5DD York
Université de Grenoble I (Université Joseph Fourier)
France
Address
385 Avenue De La Bibliothèque
38041 Grenoble